Scientific exploration, collaboration and labor division in the large language model era

Large language models (LLMs) have entered scientific workflows rapidly, but their diffusion may be associated with changes extending beyond writing and productivity. We link PubMed Central full text to OpenAlex publication and collaboration histories for 775,323 biomedical and biomedical-adjacent researchers and analyze CRediT statements from 137,120 multi-author papers. After 2022, sampled researchers published across more and more distant fields, entered fields in which they had not previously worked, and collaborated with more interdisciplinary sets of researchers. These changes were strongest among established researchers and those affiliated with non-English-speaking low- and middle-income countries. Authors with stronger AI-writing signals were already more interdisciplinary and exploratory before widespread LLM adoption, and the gaps relative to matched authors with weaker signals widened afterward. Yet their research interdisciplinarity was less closely associated with collaborator diversity. Team contributions also became more differentiated: contributors reported narrower role sets, shared fewer roles with coauthors, and formed less rigid role bundles. The evidence is descriptive rather than causal and documents a set of post-2022 changes in scientific exploration, collaboration, and reported labor division.